Neural networks based immune recognition model for aeroengine coupled rotor system fault detection

Aeroengine is a high-speed coupled dual-rotor mechanical system. Some of its abnormal vibration phenomena cannot be explained by linear theory. Therefore, the study on how to detect its fault is of great worth. In this paper, a neural network detector is constructed based on the principle of immune recognition, and the abnormal characteristics of engine rotor vibration signal are captured by the detector. The abnormal vibration signal features are trained and stored in the detector. If the signal samples to be detected match what have been stored in the detector, the exact fault will be found. The vibration experiments show that this method has high sensitivity for the detection of coupled dual-rotor system, and can effectively identify the common faults of aero-engine rotor system.